I'm attempting to learn to build from github, and have followed hashcode's directions on the KFire-Android local manifest for jb-CM10. I've even gone back and verified I have initialized and set up my build environment correctly. I am having problems syncing the repo's. I get to 38%, and get a "fatal: Unable to look up github.com (port 9418) (Name or service not known)" error. Could this be a firewall issue?
Related
Okay, so recently I got a wild hair to try compiling CM on my MBP, and proceeded to pull and install all of the various items described by the CM wiki here:
http://wiki.cyanogen...dible_(Mac)
I realize that it says the wiki hasn't been tested, and that it's for CM6 and OSX 10.6.5, but I figured / hoped that not too much had changed... which is apparently wrong. So here's what happens: I've got everything installed correctly as far as I can tell- Java for Mac update 4 (most current), the most current MacPorts (1.9.2), I've built a little disc image workspace (15gb, as suggested by the wiki). I've pulled down the repo, ADB'd the proprietary files from the Inc, etc. etc., all with success (i.e. no error messages and it looks clean).
Everything LOOKS okay, right? I should be able to just build this puppy and keep it movin', right? Wrong.
I follow all of the wiki instructions *exactly*, the build runs for about 20+ minutes, gives me an assertion error, and then returns me to command line (no .zip file created). No other errors anywhere that I can see- lots and lots of warnings (is that normal?) but no ERRORs. So, I figure maybe I'll try these boards and see if someone with more experience in this area has any suggestions. I briefly consulted Slayher about it and he was saying that he remembers something about Macs and "elf.h" not playing nice with Linux kernels... but things I've found by Googling say that those issues SHOULD be corrected by the most recent MacPorts.
So, again, I've followed the wiki instructions to a "T"... Twice, if you can believe that- I thought maybe I'd done something wrong the first time so I trashed my workspace and started over from scratch. Same result, the compile runs, looks the same, and then there's no "finishing message" or .zip in the /inc folder. Anyway, here is a link to the pastebin of the compile, if anyone wanted to take a look. Directly after the last line that is posted to this pastebin, it returned me to command line- I just didn't post my command line for privacy's sake. If you look at this link, line 2496 (and those above and below it) look to be of particular interest, but what do I know...
http://pastebin.com/CFRyh8ue
Anyone have any thoughts on compiling CM7 with a Mac running 10.6.6/7? Any help would be much appreciated. I'm working on getting access to an Ubuntu laptop, but again I'd really like to be able to do this on my MBP if I can.
Thanks for reading.
-x-
**EDIT: I've tried looking for help on this, including (as I mentioned) some pretty learned people (Slayher, CUViper, etc.) but I don't want to continually bug them... and my google / XDA searches haven't come up with anything so far. Again, any help would be appreciated... there's gotta be someone out there compiling CM7 on a Mac, right? Maybe?
I am slowly trying to figure out how to create my own nightlies for AOKP and i have my environment set up, but i am still getting some errors.
I was hoping someone could look into the following pastebin entry for me.
www.pastebin.com/6hZkVEqr
I noticed something about folder src not found.
If anyone could help me out that would be great.
To anyone who has been successfully building, I cant figure out this error. I tried syncing with TheMuppets repo and using the extract files on my phone, but I keep getting this error
Code:
make: *** No rule to make target `vendor/htc/m8/proprietary/vendor/lib/libtime_genoff.so', needed by `/home/keithn/cm11/out/target/product/m8/obj/lib/libtime_genoff.so'. Stop.
Any assistance would be apreciated.
Looks to me like you haven't set up your environment properly.
Did you get it fixed.
So I am attempting to build Bean Stalk for the M8 and I am getting the same build error.
I am using Builduntu latest release, should be setup to build out of the box.
http://soupdawg.wordpress.com/2014/03/06/builduntu-install-disc-android-development-os/
Code:
make: *** No rule to make target `/home/pimpmaneaton/BS/out/target/product/m8/obj/lib/libtime_genoff.so', needed by `/home/pimpmaneaton/BS/out/target/product/m8/obj/SHARED_LIBRARIES/libandroid_runtime_intermediates/LINKED/libandroid_runtime.so'. Stop.
I am getting the same error and have the OS updated and also gone through many of the guides that show how to set up a Ubuntu 14.04 machine up to build.
Hoping that someone might be able to shed some light on this. Thanks in advance.
EDIT: Got it figured out.........I forgot to sync vendor/htc.
You need to sync the blobs https://github.com/TheMuppets/proprietary_vendor_htc
ashyx said:
Looks to me like you haven't set up your environment properly.
Click to expand...
Click to collapse
Nothing to do with the environment. It states the issue clearly. It's missing proprietaries.
Yup you guys got it. I had them synced to the wrong location which is why I was confused about it.
Sent from my SHIELD Tablet
Hello,
I'm building this ROM for my device and at the stage of syncing the source code, I encountered two errors which made my sync exit. They were:
1) gnutls_handshake() failed: A TLS packet with unexpected length was received.
2) unable to access source.google.com
--
I know this is the wrong place for posting this, but I haven't got good help over the internet so I tried to seek some help in here.
I'm on Ubuntu 14.04 LTS.
Hi all,
Since I haven't seen a kernel tree on XDA that had the mediatek drivers (the ones xiaomi has put in a separate git repo), I thought I'll put them back into the tree.
The drivers have been taken out of the tree from drivers/misc/mediatek. To make them work at all, the makefiles need some changes.
1) In the connectivity folder, to build more than just the wifi drivers, the others folders must be added back into the build (at least, obj-y += {folders}/).
2) The include directives must be changed to include from $(srctree)/drivers/misc/mediatek/... instead of $(TOP)/vendor/.../kernel_modules/...
Besides that, there are some minor changes here and there to get them working properly.
I've managed to get it to boot correctly after disabling stock modules (by moving the folder, otherwise they seem to get inserted even if the drivers are already in the kernel). However, even if the connac device appears in the /sys/devices/platform, wifi still fails to turn on. GNSS turns on, but it is very slow. Only some gps and glonass sats show up, much fewer than normally should, and I could not get a fix (this might be either because there's some firmware problem preventing beidou and galileo from showing up, or the system is too slow and hasn't yet loaded beidou and galileo data - can't tell because I've been indoors and even on stock kernel, cold TTFF is quite long).
Care to share some thoughts? Some resources, maybe?
P.S.: I've not published the merged tree because I've poked it a lot and it's a mess.
I've dug more into this. It's quite likely that the drivers released by Xiaomi don't match the shipped kernel (or the hardware on the device).
It looks like the released wifi driver has trouble accessing the hardware registers and ports on the wifi block. Also, the firmware binary specified in the driver source doesn't match those on vendor (it's searching for WIFI_RAM_CODE_soc1_0_1_1, while stock uses WIFI_RAM_CODE_soc1_0_2a_1; these are hardcoded in the driver). I've searched through the sources for the file loaded by the stock kernel and it's not mentioned anywhere in the source.
Edit: Actually it does match, it's the flavoured firmware branch. I was confused by the comments above the code.
Edit2: Everything boils down to a missing definition in the Makefiles/build options. See below.
Ok, finally found the problem. The modules need to be built with WIFI_IP_SET=2. This flag changes both the name of the firmware and some padding in the structures. That's why it failed so hard.
AAAArgh...
If this works ok, I'll try to post a working tree next couple of days.
I've put the sources on github. Wifi, GPS, NFC, Bluetooth work built-in.
That said, there still seems to be some problems (I'm seeing reboots every 15 mins or so).
I know for a thing that there's a problem with the embedded secure element (somewhere along the TEE/NFC stack), and I think that's what causes the reboots.
Couldn't get the eSE to work; if anyone's interested, the error looks like this:
Code:
[ 197.239862] -(0)[240:teei_switch_thr][TZ_LOG] ESE-S | [phNxpEsePal_spi.cc:171/spiread_dma] <err> dma read nad over time.\x0d
[ 197.241310] -(0)[240:teei_switch_thr][TZ_LOG] ESE-S | \x0d
Which sounds like the eSE server in the TEE can't talk properly to the eSE chip over the spi bus.
However, I've disabled the secure element in Android, and it seems like the problems are gone (though secure card wallets, like Google Pay will have to use a software implementation).
I've disabled the secure element in Android by (replace removal by moving to a backup location at your discretion)
Code:
rm /vendor/etc/init/[email protected]
rm /vendor/etc/init/[email protected]
rm /bin/hw/[email protected]
rm /bin/hw/[email protected]
pm hide com.android.se
It might be that only the last command is necessary though (if you do test this, please reply - I only went up on the chain of dependencies).